[mikestar]

I've actually played a fair amount of rubber back in the days. I must be getting old--these days most social games seem to be Chicago (4 deal).

Not many duplicate players grasp how rubber bridge scoring affect tactics. The typical duplicate player plays rubber like he's playing IMPs and will do OK but get at lot of his close decisions wrong.

By the way, a very fair number of rubber bridge players when the game was in its heyday were quite ignorant of proper tactics.


Let's look at some numbers. What is the value of game? Winning the first game of the rubber is worth 350 (not the 300 given for a game in an unfinished rubber, which was the origin of 300 for game NV in duplicate). The reasoning: You have a 50% chance of winning the second game for a 700 rubber, for an expectation of +350. If you lose the second game, you have a 50-50 chance of winning or losing a 500 rubber, for an expectation of 0. Value of the first game: 350.

The second game of the rubber is also worth 350. If the side that won the first game wins the second game, they change their 350 expectation into a 700 score=+350. If the side which lost the first game wins the second game, they wipe out the other sides 350 advantage=+350.

The third game is worth the 500 rubber score.

So for prempting/sacrificing, we can see some differences vs. IMPs. It would seem that we can preempt just a tiny bid more aggresively at neither vul, as their 4M is worth 470 instead of 420, but the difference is slight and is outweiged by the absense of the IMP scale reducing the magnitude of our big losses--so IMP tactics should work here. There is a big difference at favorable: their 4M is worth 470 instead of 620--we should bid as if neither vul. Again we can can follow IMP standards at both vul and unfavorable: at both vul, theier 4M is 620 same as at duplicate, at unfavorable their 4M is 470 rather than 420, but the disasters hurt more.

The value of a very small partscore (<40) is less, but these are rare in a good game.


Now where the difference in the games really show up is in partscores. A partscore is worth somewhere between 1/4 and 1/3 of the value of the game. Partscores get converted about half the time, but the partscore was only an advantage if you convert it when yoiu couldn't have made game anyway. So let's say a partscore is worth 150 with both vul and 100 otherwise.

Constructive game bidding needs to be more conservative than IMPs, more like MP.

Competitive partscore bidding with nothing on the score will be very similar to IMPs, maybe even more aggressive as the parscore is so much more valubale than in duplicate: letting them make 3♥ when we can make 3♠ is a 380 point swing. Where everbody goes wrong is when some one side or the other has a partscore.

If both sides have partscores, competive bidding should be very agressive notwithstanding the risks. Now letting them make 3♥ when we can make 3♠ is a double game swing.

If our opponents have a partscore, we should go quietly. Thier game is of less value as they already have an expectation of 100 or 150. So giving up 300 to stop 2M = 310 or 360 is a small gain and the time the sac is too expensive are more costly.

If we have a partscore, we should be ready with the axe for the same reason: our game is less valuable but the penalties are the same size.

If our opponents have a partscore, we should bid game freely. Say both vul they have 60 on and we bid and make 4M. Our gain is 120 for the trick score + 500 for the game + 150 for wiping out their expecation = 770, while the loss is 90 for tricks if we had stopped in 3 + 150 for partscore + 100 for the undertrick =340. 31% is enough if we don't allow for doubles or down more than one. Rather less than the % needed vul at IMPs.

For slam bidding, if both sides have a partscore there is no effect. If only our side has a score, we can live it up. Our slam is just as valuable but the game we blow if we go down is worth 100/150 less, because we still keep our partscore.

If only they have a partscore, we have to tighen up: the game we blow if we go down is worth 100/150 more for wiping out the enemy partscore advantage.

This post has been edited by mikestar: 2005-October-13, 01:52

[Blofeld]

Nice post!

Everything you put forward is fairly straightforward and logical, but I had never actually worked out the numbers or considered it in this much detail before.

Only thing I don't follow: why is their game worth 670 rather than 620 at game all?

[Trpltrbl]

mikestar, on Oct 12 2005, 10:05 AM, said:

I've actually played a fair amount of rubber back in the days. I must be getting old--these days most social games seem to be Chicago (4 deal).

Chicago

There are several versions of this game, also known in the official rules as Four-Deal Bridge. As this name suggests it is a game for four players which is complete in four deals, unlike Rubber Bridge, where the length of a rubber is indefinite. This greater predictability has made it popular in some American clubs where Rubber was formerly played.


The vulnerability varies from hand to hand in a fixed pattern as follows:


Hand 1: Dealer North; neither side vulnerable
Hand 2: Dealer East; North-South vulnerable
Hand 3: Dealer South; East-West vulnerable
Hand 4: Dealer West; both sides vulnerable

But there is even a more Chicago style

Chicago with Duplicate Scoring

Chicago is sometimes played using duplicate scoring. There is no accumulation of part scores or games from deal to deal - each deal is scored separately, and a team making a part score gets an immediate bonus of 50 as in duplicate. The sequence of vulnerability is fixed as in the standard version.


A multiple of four hands can be played, repeating the sequence of vulnerabilities as often as necessary. The result is simply the total score over the deals played.

And last but not least and very few people play this style.

Chicago with Russian Scoring

The following method of scoring Chicago originated in Russia. It eliminates some of the luck of the deal by introducing an element of IMPs scoring.


On each deal, there is a target score which depends on the number of high card points held. The cards are played in front of the players, as in duplicate. At the end of the play, the high card points held by each side are counted, according to the following scale:

each ace: 4 points
each king: 3 points
each queen: 2 points
each jack: 1 point
There are 40 points in all. The team which held more high card points finds its target score, which depends on whether they were vulnerable or not, from the following table:

High Card Target
Points Not Vul. Vulnerable
20 0 0
21 50 50
22 70 70
23 110 110
24 200 290
25 300 440
26 350 520
27 400 600
28 430 630
29 460 660
30 490 690
31 600 900
32 700 1050
33 900 1350
34 1000 1500
35 1100 1650
36 1200 1800
37 1300 1950
38 1300 1950
39 1300 1950
40 1300 1950

The difference between the target score from the above table and the actual score is then converted to IMPs, using the standard IMP table. The total IMP scores over a series of hands are totaled to give an overall result.

For example, suppose we are East-West, and on the second deal of a Chicago we bid three hearts and make 10 tricks. We then count our high card points and discover that between us we had 24. We were vulnerable, so our target score from the table was 290. We actually scored 170 (90 for the contract plus 30 for the overtrick plus 50 for the part score). So we are 120 points short of our target. Therefore using the IMP table, our score for this hand is minus 3 IMPs .

GBB

[han]

The least random form of bridge must be par bridge. Here is how I think 4-people bridge should be played:

After the deal is over, the hand is played double dummy (so with open cards) but with the hands reversed (so EW now play the NS hands and NS the EW hands). In theory, the par should be reached this way, except that on some hands people might not find the best double dummy play and contract. Depending on your tastes you can agree that partners may or may not talk during the replay.

[barmar]

Don't forget one of the other scoring differences in rubber: honors. In a suit contract, any player who holds 4 of the top 5 cards in the trump suit gets a 100-point bonus, and if he holds all 5 top cards he gets 150 points. In a notrump contract, 150 honors is scored for holding all the aces.

When non-vulnerable, 150 honors is close to the value of an additional doubled undertrick, so you can often preempt a level higher if you have a solid suit.

[Blofeld]

Hannie, on Oct 12 2005, 02:29 PM, said:

The least random form of bridge must be par bridge. Here is how I think 4-people bridge should be played:

After the deal is over, the hand is played double dummy (so with open cards) but with the hands reversed (so EW now play the NS hands and NS the EW hands). In theory, the par should be reached this way, except that on some hands people might not find the best double dummy play and contract. Depending on your tastes you can agree that partners may or may not talk during the replay.

I'd devised the same method of play.

I've never actually tried it, though I think it's an excellent idea.

[mikestar]

Blofeld, on Oct 12 2005, 06:33 PM, said:

Nice post!

Everything you put forward is fairly straightforward and logical, but I had never actually worked out the numbers or considered it in this much detail before.

Only thing I don't follow: why is their game worth 670 rather than 620 at game all?

My bad I was writing too fast--I have corrected the post.

[mikestar]

barmar, on Oct 12 2005, 08:01 PM, said:

Don't forget one of the other scoring differences in rubber: honors. In a suit contract, any player who holds 4 of the top 5 cards in the trump suit gets a 100-point bonus, and if he holds all 5 top cards he gets 150 points. In a notrump contract, 150 honors is scored for holding all the aces.

When non-vulnerable, 150 honors is close to the value of an additional doubled undertrick, so you can often preempt a level higher if you have a solid suit.

Quite correct. holding honors can also tilt the odds just a bit in hihg level decisions such as whether to bid on over a sacrifice: they increase the reward of being right and reduce the cost of being wrong--especially if the sacrificing side might also have honors.

[kfgauss]

Hannie, on Oct 12 2005, 07:29 PM, said:

The least random form of bridge must be par bridge. Here is how I think 4-people bridge should be played:

After the deal is over, the hand is played double dummy (so with open cards) but with the hands reversed (so EW now play the NS hands and NS the EW hands). In theory, the par should be reached this way, except that on some hands people might not find the best double dummy play and contract. Depending on your tastes you can agree that partners may or may not talk during the replay.

A less ideal but much quicker method is to give extra points based on HCP. The way I've seen this done is to give the side with fewer HCP 60 points for each HCP they fall short of 20. For example, if your side has 26 HCP, your opponents will get 360 points above the line on that hand.

Andy